Pipelined Cooperative Spectrum Sensing in Cognitive Radio Networks

Cooperation can improve the performance of spectrum sensing. However, the sensing overhead is generally increasing with the number of cooperating users as more data needs to be reported to the fusion center. Most existing works assume a general time frame structure in which spectrum observing and sensing results reporting are conducted sequentially. We argue that this frame structure is inefficient, since the time consumed by reporting contributes little to the performance of spectrum sensing. In this paper, we propose a pipelined spectrum sensing framework, in which spectrum observing is conducted concurrently with results reporting in a pipelined way. By making use of the reporting time for sensing, the new framework provides a much wider observing window for spectrum measurement, which results in a performance improvement of spectrum sensing. Besides, we also present a multi-threaded sequential probability ratio test method (MTSPRT) which is very suitable for the pipelined framework as the data fusion technique. The MTSPRT method can improve the sensing speed significantly. Numerical results indicate that our pipelined sensing scheme incorporating with MTSPRT shows a better performance than the cooperative sensing based on the general frame structure.